I wired the machine today. All the network cable I had was solid core, so I used 22ga hook up wire instead. It’s not shielded wire, but so far so good. I decided to leave the machine configured for dual end stops even though I have not wired them yet. All I need now is a 40mm fan for the RAMPS case.
I gave the machine a good workout today. The crown got a bit boring so I wrote a program to dither photographs using lines at various angles. My plan was to save the line segments as an SVG file and then generate gcode with Estlcam, but the time it took for the software to generate the paths was unbearable. Maybe it has something to do with running it under wine on linux. Instead of waiting for Estlcam, I programmed gcode output directly into my dithering software and even added a crude path merging and optimization scheme to reduce the time spent lifting the pen and moving around without drawing. The attached picture took 56 minutes to draw. I used six gray values and an ultra fine tip sharpie after initial attempts with ball point pens failed. They would not draw consistently. The sharpie, on the other hand, tends to bleed a little and is a bit thicker than I would like, so I had to reduce the gray value resolution. I have some ideas how to improve the result and calibrate the dithered gray values to obtain more contrast, but I am pretty happy with how it turned out.
In my software, I calculated time spent on drawing moves, travel without drawing and moving the pen up and down. The total of all three categories was a bit under ten percent less than the actual time, as I did not take acceleration into account. Interestingly, my calculations show that just over half the time is spent moving up and down, about five percent moving without drawing and the rest on actual drawing. And that is with a clearance plane of only 1.25mm. I used speeds of 35mm/s in-plane and 10mm/s out-of-plane. I think if one wanted to make a large drawing with this or similar techniques a fast pen lifter would help. Maybe something with a solenoid.
I am really excited about how consistent the machine is. The line spacing appears to be spot on. My path optimization scheme always tries to find the closest path endpoint, so a lot of times it would fill in a line in between two neighbors that were drawn much earlier. I cannot tell any difference, so there is no error accumulation at all. It is truly an amazing design!